We need checks and balances to ensure data-driven predictions don't become prejudices.
“Do you know why the French hate traffic cameras?” he asked me. “It’s because it makes it hard for them to cheat on their spouses.”
He contended that while it was possible for a couple to overlook subtle signs of infidelity — a brush of lipstick on a collar, a stray hair, or the smell of a man’s cologne — the hard proof of a speeding ticket given on the way to an afternoon tryst couldn’t be ignored.
Humans live in these grey areas. A 65 mph speed limit is really a suggestion; it’s up to the officers to enforce that limit. That allows for context: a reckless teen might get pulled over for going 70, but a careful driver can go 75 without incident.
But a computer that’s programmed to issue tickets to speeders doesn’t have that ambiguity. And its accusations are hard to ignore because they’re factual, rooted in hard data and numbers.
Did big data kill privacy?
With the rise of a data-driven society, it’s tempting to pronounce privacy dead. Each time we connect to a new service or network, we’re agreeing to leave a digital breadcrumb trail behind us. And increasingly, not connecting makes us social pariahs, leaving others to wonder what we have to hide.
But maybe privacy is a fiction. For millennia — before the rise of city-states — we lived in villages. Gossip, hearsay, and whisperings heard through thin-walled huts were the norm.
Shared moral values and social pressure helped groups to compete better against other groups, helping to evolve the societies and religions that dominate the world today. Humans thrive in part because of our groupish nature — which is why moral psychologist Jonathan Haidt says we’re 90% chimp and 10% bee. We might have evolved as selfish individuals, but we conquered the Earth as selfish teams.
In other words, being private is relatively new, perhaps only transient, and gossip helped us get here. Read more…
Data and context are always linked, data outputs beyond visualizations, state of the computer book market.
This week on O'Reilly: Mike Loukides explained why problems arise when data is taken out of social contexts, Robbie Allen looked at six ways insight can be extracted from datasets, and Mike Hendrickson analyzed the current state of the computer book market.